Sankhya: The Indian Journal of Statistics
2001, Volume 63, Series B, Pt. 2, pp. 199--217
WAVELET PACKET MODELLING OF INFANT SLEEP STATE USING HEART RATE DATA
GUY P. NASON University of Bristol, England THEOFANIS SAPATINAS, University of Cyprus, Cyprus
ANDREW SAWCZENKO, Instute of Child Health, Bristol, England
SUMMARY. We show how a recently developed wavelet packet modelling methodology could be useful for infant sleep state classification using heart rate data. The suggested approach produces adequate classification rates when applied to recordings taken at different ages from an infant who was put to bed at night. As well as classification, this approach gives us valuable information about the relationship between sleep state and heart rate. The statistical model tells us which sorts of wavelet packets of heart rate are most important for classifying sleep state.
AMS (1991) subject classification. 62M10.
Key words and phrases. Antedependence models; infant sleep state classification; linear discriminant analysis; variable selection; wavelets; wavelet packets
Full paper (PDF)